SalesPipeline & Forecasting

AI Prompts for Win/Loss Analysis

Win/loss analysis asks buyers why they decided the way they did — and then compares that with what the sales team believed. The gap between the two is the finding. It only works with a structured interview that lets buyers be candid (usually not run by the rep), a coding scheme that turns anecdotes into consistent reasons, and a review that changes something: the ICP, the messaging, the demo, the qualification criteria.

These prompts design the interview, code the findings, and drive the changes. They need real interviews or at least real CRM loss reasons; the model can structure and pattern-match but cannot know why your buyers chose.

Before you use these

Have these ready to replace the highlighted [variables]:

A set of recent wins and losses with deal profile, competitor, and the rep's stated reason
Access to buyer contacts for interviews (or transcripts already collected)
Your current loss-reason codes in the CRM and how reliably they are used
The decisions the analysis should inform (targeting, messaging, product, process)

The prompts

1. Design the win/loss interview guide
2. Code the findings into consistent reasons
3. Turn patterns into changes

1. Design the win/loss interview guide

Best forAn interview that buyers will answer honestly and that produces comparable findings across deals.

Inputs needed

Deal types
Hypotheses about why you win and lose
Who will interview

How to use itGive your hypotheses so the guide tests them without leading. Ask for separate guides for wins and losses and for the opening that earns candor.

Expected outputInterview guides for wins and losses with opening, question sequence, probes, the hypotheses each question tests, and the logistics that protect candor.

Act as a win/loss program lead designing interview guides for [company].

Deal profile: [segments, typical committee, competitors]
Hypotheses: [why we think we win; why we think we lose]
Interviewer: [third party / non-sales internal / sales]
Decisions the program informs: [targeting, messaging, product, process]

1. Opening (2 minutes): the framing that earns candor — purpose, confidentiality, no sales follow-up, why their view matters.
2. Loss guide: sequence from the buyer's process (how the need arose, who was involved, what criteria mattered) → the evaluation (how each vendor performed on the criteria, in their words) → the decision (what tipped it, what we could have done differently) → the relationship (how the rep, demo, proposal and pricing landed). Include probes that get past politeness ('what would have had to be true for us to win?').
3. Win guide: same sequence, focused on what nearly lost it and what the competitor did well.
4. Map each question to the hypothesis it tests; flag any question that leads.
5. Logistics: timing after decision, length, recording and consent, how the rep is kept out of the interview but informed of findings.
6. The five questions to keep if the buyer only gives ten minutes.

Questions must be open and neutral. The buyer should be describing their decision, not grading us.

2. Code the findings into consistent reasons

Best forTurning interview transcripts and CRM notes into a small, consistent set of reasons that can be counted.

Inputs needed

Interview transcripts or notes
CRM loss reasons
Coding scheme or the need for one

How to use itPaste several interviews. The model proposes or applies a coding scheme with primary and secondary reasons, quotes the evidence, and flags where the rep's reason and the buyer's differ.

Expected outputCoding scheme, coded deal table with primary/secondary reasons and quotes, rep-vs-buyer reason comparison, and the counts by segment.

You are coding win/loss findings.

Interviews: [transcripts or notes per deal, with deal profile and outcome]
CRM reasons: [the rep's stated reason per deal]
Coding scheme: [existing codes, or ask for a proposal]

1. Coding scheme (propose if none): a small set of primary reasons — e.g. fit/need mismatch, price/value, product capability, competitor strength, relationship/trust, process/timing, champion strength, proof/references, implementation risk — each with a definition and an example. Keep it under twelve.
2. Code each deal: primary reason, secondary reason, with the quote that supports each. Mark confidence.
3. Compare the buyer's reason with the rep's CRM reason per deal; classify the mismatch (rep blamed price, buyer cited fit; etc.).
4. Counts: reasons by outcome, by segment, by competitor, by deal size band — with sample sizes.
5. Surprises: reasons the hypotheses did not include; reasons that appear in wins as near-losses.
6. Quotes worth keeping verbatim for enablement.

Do not infer a reason the buyer did not give. Where the interview is ambiguous, code as 'unclear' rather than guess.

3. Turn patterns into changes

Best forDeciding what the findings should change, with owners and a way to know it worked.

Inputs needed

Coded findings
Current ICP, messaging, process
Decision owners

How to use itGive the counts and the quotes. The model links each pattern to a specific change in targeting, messaging, demo, proposal, pricing or process, and defines the metric that would show the effect.

Expected outputPattern → change table with owner, effort, expected effect, the metric and the review date; the findings that should not drive change yet; the enablement summary for reps.

Act as a revenue leader converting win/loss findings into changes.

Findings: [reasons by outcome and segment with counts; rep-vs-buyer mismatches; quotes]
Current state: [ICP, key messages, demo storyline, proposal, pricing approach, qualification criteria]
Owners: [marketing, enablement, product, sales ops, sales leadership]

1. For each pattern with sufficient sample: the change it implies (ICP criterion, message, demo moment, proposal section, pricing structure, qualification criterion, rep behavior), the owner, effort, expected effect, and the metric that would show it (win rate in a segment, loss reason frequency, stage conversion).
2. Patterns that are interesting but under-sampled: what additional data would confirm them before acting.
3. Rep-vs-buyer mismatches: the coaching or CRM change that reduces them (e.g. mandatory buyer-sourced loss reason).
4. Product findings: how to present them to product so they are prioritized (frequency, revenue affected, quotes).
5. Enablement summary for reps (one page): what buyers said, what to do differently, the quotes.
6. Review date and the next round of interviews.

Prioritize by revenue affected and confidence. Do not change the ICP on three interviews.

Worked example

A vendor whose reps recorded 70% of losses as 'price'. Eighteen buyer interviews coded price as the primary reason in four; the most common buyer reason was 'implementation risk' — the buyers did not believe the rollout timeline. The change was a proposal section on implementation with references, and a demo moment on onboarding; the reps' 'price' code dropped to 35% the following quarter. Illustrative only.

Related prompts

Logical next step

After this, most sales teams move on to Ideal Customer Profile (ICP) Definition.

Get the free B2B Sales AI Starter Kit → Nine of these prompts as a target → engage → qualify workflow with an intake worksheet, delivered by email. See what's inside

All Sales prompts · Search the full library